Back in the day, if you wanted to master a craft, you found someone who had already walked the path. Blacksmiths, artists, engineers – they all learned through mentorship. Skills weren’t just taught, they were absorbed. You’d watch, ask questions, try, fail, and try again. No shortcuts. No magic formulas. Just real learning, passed from one hand to another.
That’s exactly how large language models (LLMs) are trained. They don’t just memorize data – they go through a hardcore learning process, much like an apprentice refining their craft. With a distillation-based approach, knowledge is transferred from larger, more complex models to smaller, more efficient ones. It’s a process supervised by other models that act as mentors, refining responses, filtering out errors, and reinforcing patterns. Just like the old masters, LLMs don’t achieve expertise overnight—it’s earned through countless cycles of training, feedback, and refinement.
Then machines showed up and started replacing hands. Software came along and wiped out clerical work. With AI writing code, answering emails, and even generating entire apps. So, does that mean mentorship is dead?
AI Is a Beast, But It’s Not a Teacher
The speed at which LLM spits out code is something we’ve never seen before. A prompt, a click, and boom – lines of code are here. It feels powerful. And for someone who doesn’t know what they’re looking at, it seems like magic.
But magic tricks only work when you don’t know how they’re done.
Any sufficiently advanced technology is indistinguishable from magic.
Arthur C. Clarke
The thing is, AI doesn’t understand code – it predicts it. It’s like an overconfident intern who talks a big game but has never actually deployed anything to production. If you don’t know how to code yourself, AI-generated code can be a total disaster. It might work… or it might introduce security holes, fail under real-world conditions, or just break everything. And you won’t even know why.
So if you think AI is a shortcut to becoming a great developer… no, it’s not.
AI + Knowledge = Superpower
Now, here’s the flip side. If you do know how to code, AI becomes a cheat code. It’s like giving a seasoned blacksmith a power hammer instead of a regular one. They still know the craft, they still shape the metal, but now they can do it way faster.
I remember about 10 years ago when I was writing mundane code to extend a system, I was dreaming about the day I could delegate that kind of work to AI agents. That’s exactly what’s happening. AI already writes 90% of the code for me. But it’s not just about letting AI take over – it’s about knowing how to guide it.
An experienced dev using AI isn’t just generating code; they’re sculpting it. They know when AI is bullshitting, they know how to tweak prompts to get better results, and they know how to take AI’s output and make it actually useful. You still need to be careful, double-check everything, and sometimes break down large tasks into smaller steps to get the right implementation. It’s about fully controlling the process, not just accepting whatever AI spits out.
This is the difference between someone who just copies AI-generated code and someone who shapes it into something real.
AI Won’t Make You Great, But It’ll Make Great Devs Faster
A junior dev who doesn’t understand what they’re doing will still struggle, AI or not. They might move a little faster, but they won’t truly learn. They won’t get why things break, they won’t be able to debug properly, and when things go south, they’ll be stuck.
But someone who already understands coding fundamentals? Who knows how to structure software, debug, and optimize performance? AI takes them from fast to insanely fast. They don’t spend hours writing boilerplate. They focus on refining, testing, and making sure everything actually works.
And the ones who get really good at prompting AI? The ones who can guide it, correct its mistakes, and use it like an actual tool instead of a crutch? They’ll be so ahead.
The Real Shortcut Is Still Learning
AI is a tool. A powerful one. But it won’t do the work for you.
If you’re serious about getting good at this – if you want to build code that’s solid, safe, and actually does what it’s supposed to – you still have to put in the work. AI won’t replace learning. It won’t replace mentorship. It won’t replace experience.
What it will do is make good developers unstoppable.
And if you’re just starting out? The best way to use AI is alongside real-world learning – open-source projects, mentorship, hands-on practice. AI can help speed things up, but at the end of the day, your skills will still be the thing that sets you apart.
So don’t use AI as a shortcut. Use it as a multiplier.
